Transferability of regional permafrost disturbance susceptibility modelling using generalized linear and generalized additive models

Abstract To effectively assess and mitigate risk of permafrost disturbance, disturbance-prone areas can be predicted through the application of susceptibility models. In this study we developed regional susceptibility models for permafrost disturbances using a field disturbance inventory to test the transferability of the model to a broader region in the Canadian High Arctic. Resulting maps of susceptibility were then used to explore the effect of terrain variables on the occurrence of disturbances within this region. To account for a large range of landscape characteristics, the model was calibrated using two locations: Sabine Peninsula, Melville Island, NU, and Fosheim Peninsula, Ellesmere Island, NU. Spatial patterns of disturbance were predicted with a generalized linear model (GLM) and generalized additive model (GAM), each calibrated using disturbed and randomized undisturbed locations from both locations and GIS-derived terrain predictor variables including slope, potential incoming solar radiation, wetness index, topographic position index, elevation, and distance to water. Each model was validated for the Sabine and Fosheim Peninsulas using independent data sets while the transferability of the model to an independent site was assessed at Cape Bounty, Melville Island, NU. The regional GLM and GAM validated well for both calibration sites (Sabine and Fosheim) with the area under the receiver operating curves (AUROC) > 0.79. Both models were applied directly to Cape Bounty without calibration and validated equally with AUROC's of 0.76; however, each model predicted disturbed and undisturbed samples differently. Additionally, the sensitivity of the transferred model was assessed using data sets with different sample sizes. Results indicated that models based on larger sample sizes transferred more consistently and captured the variability within the terrain attributes in the respective study areas. Terrain attributes associated with the initiation of disturbances were similar regardless of the location. Disturbances commonly occurred on slopes between 4 and 15°, below Holocene marine limit, and in areas with low potential incoming solar radiation.

[1]  C. Wright,et al.  Improved wetland remote sensing in Yellowstone National Park using classification trees to combine TM imagery and ancillary environmental data , 2007 .

[2]  Catherine H. Graham,et al.  Factors Influencing Movement Patterns of Keel‐Billed Toucans in a Fragmented Tropical Landscape in Southern Mexico , 2001 .

[3]  Antoine Guisan,et al.  Are niche-based species distribution models transferable in space? , 2006 .

[4]  David R. B. Stockwell,et al.  Effects of sample size on accuracy of species distribution models , 2002 .

[5]  M. Woo,et al.  Hydrology of Two Slopes in Subarctic Yukon, Canada , 1999 .

[6]  David R. B. Stockwell.  Generic predictive systems: An empirical evaluation using the learning base system (LBS) , 1997 .

[7]  P. Treitz,et al.  Terrain Controls and Landscape‐Scale Susceptibility Modelling of Active‐Layer Detachments, Sabine Peninsula, Melville Island, Nunavut , 2017 .

[8]  A. Brenning.  Spatial prediction models for landslide hazards: review, comparison and evaluation , 2005 .

[9]  François Charbonneau,et al.  Surface roughness estimation from RADARSAT-2 data in a High Arctic environment , 2014, Int. J. Appl. Earth Obs. Geoinformation.

[10]  M. Leibman.  Cryogenic landslides on the Yamal Peninsula, Russia: Preliminary observations , 1995 .

[11]  Wayne H. Pollard,et al.  Fifty years of coastal erosion and retrogressive thaw slump activity on Herschel Island, southern Beaufort Sea, Yukon Territory, Canada , 2008 .

[12]  C. F. Lee,et al.  Landslide characteristics and, slope instability modeling using GIS, Lantau Island, Hong Kong , 2002 .

[13]  Denis Lacelle,et al.  Climatic and geomorphic factors affecting contemporary (1950–2004) activity of retrogressive thaw slumps on the Aklavik Plateau, Richardson Mountains, NWT, Canada , 2010 .

[14]  Boris Schröder,et al.  How can statistical models help to determine driving factors of landslides , 2012 .

[15]  C. Chatfield.  Model uncertainty, data mining and statistical inference , 1995 .

[16]  Ming-ko Woo,et al.  Influence of Local Topography, Soils, and Vegetation on Microclimate and Hydrology at a High Arctic Site, Ellesmere Island, Canada , 1997 .

[17]  Kenneth M. Hinkel,et al.  The transient layer: implications for geocryology and climate‐change science , 2005 .

[18]  A. Lewkowicz,et al.  Morphology and geotechnique of active-layer detachment failures in discontinuous and continuous permafrost, northern Canada , 2005 .

[19]  B. Etzelmüller,et al.  Transferability of geomorphological distribution models: Evaluation using solifluction features in subarctic and Arctic regions , 2014 .

[20]  A. Fielding,et al.  Testing the Generality of Bird‐Habitat Models , 1995 .

[21]  P. Reichenbach,et al.  Gis Technology in Mapping Landslide Hazard , 1995 .

[22]  Fiona Marianne Gregory.  Biophysical remote sensing and terrestrial CO2 exchange at Cape Bounty, Melville Island , 2012 .

[23]  L. Dredge,et al.  Terrain characterization and evaluation : an example from eastern Melville Island , 1977 .

[24]  A. Brenning,et al.  Assessing the quality of landslide susceptibility maps – case study Lower Austria , 2014 .

[25]  C. Burn,et al.  CANADIAN LANDFORM EXAMPLES ‐ 17 RETROGRESSIVE THAW SLUMPS , 1990 .

[26]  Fujun Niu,et al.  Thaw-induced slope failures and susceptibility mapping in permafrost regions of the Qinghai–Tibet Engineering Corridor, China , 2014, Natural Hazards.

[27]  M. Torre Jorgenson,et al.  Abrupt increase in permafrost degradation in Arctic Alaska , 2006 .

[28]  T. M. Smith,et al.  A new model for the continuum concept , 1989 .

[29]  Andreas Kääb,et al.  Mountain permafrost distribution modelling using a multi‐criteria approach in the Hövsgöl area, northern Mongolia , 2006 .

[30]  Hugh M. French,et al.  The Periglacial Environment , 1977 .

[31]  A. Lewkowicz,et al.  Salinization of Permafrost Terrain Due to Natural Geomorphic Disturbance, Fosheim Peninsula, Ellesmere Island , 1999 .

[32]  P. Mclaren,et al.  Holocene Emergence of the South and East Coasts of Melville Island, Queen Elizabeth Islands, Northwest Territories, Canada , 1978 .

[33]  Oleg A. Anisimov,et al.  Climate Change and Hazard Zonation in the Circum-Arctic Permafrost Regions , 2002 .

[34]  Netra R. Regmi,et al.  Assessing susceptibility to landslides: Using models to understand observed changes in slopes , 2010 .

[35]  R. Real,et al.  Transferability of environmental favourability models in geographic space : The case of the Iberian desman (Galemys pyrenaicus) in Portugal and Spain , 2009 .

[36]  K. Beven,et al.  A physically based, variable contributing area model of basin hydrology , 1979 .

[37]  M. Luoto,et al.  Novel theoretical insights into geomorphic process–environment relationships using simulated response curves , 2011 .

[38]  P. Atkinson,et al.  Generalised linear modelling of susceptibility to landsliding in the Central Apennines, Italy , 1998 .

[39]  Alexander Brenning,et al.  Evaluating machine learning and statistical prediction techniques for landslide susceptibility modeling , 2015, Comput. Geosci..

[40]  T. Bell.  The last glaciation and sea level history of Fosheim Peninsula, Ellesmere Island, Canadian High Arctic , 1996 .

[41]  S. Lamoureux,et al.  Fluvial Impact of Extensive Active Layer Detachments, Cape Bounty, Melville Island, Canada , 2009 .

[42]  Steven J. Phillips,et al.  WHAT MATTERS FOR PREDICTING THE OCCURRENCES OF TREES: TECHNIQUES, DATA, OR SPECIES' CHARACTERISTICS? , 2007 .

[43]  A. Lewkowicz.  Dynamics of active‐layer detachment failures, Fosheim Peninsula, Ellesmere Island, Nunavut, Canada , 2007 .

[44]  B. Alt,et al.  Regional Congruence of Vegetation and Summer Climate Patterns in the Queen Elizabeth Islands, Northwest Territories, Canada , 1989 .

[45]  H. R. Payne,et al.  22. Development of a methodology for landslip potential mapping in the Rhondda Valley , 1991 .

[46]  R. Sidle,et al.  A distributed slope stability model for steep forested basins , 1995 .

[47]  R. Tibshirani,et al.  Generalized additive models for medical research , 1986, Statistical methods in medical research.

[48]  Trevor Hastie,et al.  Generalized linear and generalized additive models in studies of species distributions: setting the scene , 2002 .

[49]  C. F. Lee,et al.  Assessment of landslide susceptibility on the natural terrain of Lantau Island, Hong Kong , 2001 .

[50]  A. Brenning,et al.  Integrating physical and empirical landslide susceptibility models using generalized additive models , 2011 .

[51]  Adam Collingwood,et al.  Identifying permafrost slope disturbance using multi-temporal optical satellite images and change detection techniques , 2013 .

[52]  Trevor C. Lantz,et al.  Origin and polycyclic behaviour of tundra thaw slumps, Mackenzie Delta region, Northwest Territories, Canada , 2009 .

[53]  Douglas A. Hodgson,et al.  Quaternary geology of central Melville Island, Northwest Territories , 1984 .

[54]  Guido Grosse,et al.  Vulnerability of high‐latitude soil organic carbon in North America to disturbance , 2011 .

[55]  M. T. Jorgenson,et al.  Advances in Thermokarst Research , 2013 .

[56]  Antoine Guisan,et al.  Predictive habitat distribution models in ecology , 2000 .

[57]  Trevor C. Lantz,et al.  Increasing rates of retrogressive thaw slump activity in the Mackenzie Delta region, N.W.T., Canada , 2008 .

[58]  S. Lamoureux,et al.  Thermal Perturbation and Rainfall Runoff have Greater Impact on Seasonal Solute Loads than Physical Disturbance of the Active Layer , 2013 .

[59]  Mathieu Marmion,et al.  Effects of sample size on the accuracy of geomorphological models , 2008 .

[60]  M. Symonds,et al.  A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion , 2010, Behavioral Ecology and Sociobiology.

[61]  M. Becht,et al.  Sample size matters: investigating the effect of sample size on a logistic regression susceptibility model for debris flows , 2014 .

[62]  Julian D. Olden,et al.  Assessing transferability of ecological models: an underappreciated aspect of statistical validation , 2012 .

[63]  T. Bell,et al.  MASSIVE ICE FORMATION IN THE EUREKA SOUND LOWLANDS: A LANDSCAPE MODEL , 1998 .

[64]  S. L. Kuriakose,et al.  Spatial data for landslide susceptibility, hazard, and vulnerability assessment: An overview , 2008 .

[65]  K. Sassa,et al.  Landslide Science for a Safer Geoenvironment: Volume 3: Targeted Landslides , 2014 .

[66]  Isik Yilmaz,et al.  Landslide susceptibility mapping using frequency ratio, logistic regression, artificial neural networks and their comparison: A case study from Kat landslides (Tokat - Turkey) , 2009, Comput. Geosci..

[67]  Robert H. Fraser,et al.  Distribution and growth of thaw slumps in the Richardson Mountains-Peel Plateau region, northwestern Canada , 2015 .

[68]  J. Neter,et al.  Applied Linear Regression Models , 1983 .

[69]  S. Manel,et al.  Evaluating presence-absence models in ecology: the need to account for prevalence , 2001 .

[70]  Tom Fawcett,et al.  An introduction to ROC analysis , 2006, Pattern Recognit. Lett..

[71]  Anthony Lehmann,et al.  GRASP: generalized regression analysis and spatial prediction , 2002 .

[72]  Ulf Bergström,et al.  Transferability of predictive fish distribution models in two coastal systems , 2009 .

[73]  C. Burn,et al.  Near‐surface ground ice in sediments of the Mackenzie Delta, Northwest Territories, Canada , 2005 .

[74]  Edward C McRoberts,et al.  The Stability of Thawing Slopes , 1974 .